Mike Frysinger wrote:
that certainly sounds reasonable to me. if the stable cant be maintained, let
the common workflow of developers transition it back to ~arch until someone
has the time to keep arch usable. changing profiles.desc accordingly should
be done ahead of time. perhaps a new category for profiles.desc ? "exp" for
such ports ? i could see all *-fbsd ports being moved there. tweak repoman
to be less verbose about dep issues for such profiles and we're set.
Sounds like a plan. 'exp' would be the 'status' field? I need to remove
2006.1, as that profile has been a big holdup due to it not being glibc-2.4
friendly (or one of the newer glibcs back in that era; I forget). Even
pondering just outright booting 2007.0, as I've been using 2007.1-dev since I
commited it long ago, and haven't had an issue with it really. I can then put
2008.0-dev together and use it as a launch platform for ~arch migration.
i see dropping keywords as a very last resort. getting a port *back* into the
tree is a *tremendous* amount of work (i went through it and it was hell),
while keeping ~arch alive is a sliver of effort and generally not a blocker
for package maintainers.
Aye, I believe that was sh's removal and subsequent re-add?
Part of the hangup lately has been our kernel support. O2 systems are dead in
the water in 2.6.24, and only work in 2.6.23 if you apply a hack to serial_core
(a hack that only masks a problem rather than fixes it). Octane's I can still
forward port, but with the upstream author having moved onto other interests, if
something breaks badly enough from one version to the next, then I run the risk
of getting stuck on a particular version permanently.
Indigo2 R10000's may wind up getting resurrected, as support for that is
actually headed into upstream now, so it'll be the end of patching for that
system. Though the gcc patch needs fixing.
And I'm really considering dropping our mips3 (Indigo2/Indy R4x00) support to
cut back on the number of stages and netboots pumped out (-3 and -1,
respectively, when they get pumped out). R4x00 is an odd CPU, with a ton of
variations, and of them, only the R4400 ever seems to work well at all.
The hard part is finding time and motivation. My attention span lately has been
worse than a goldfish's. That said, however, profiles should be doable come the
weekend, at least for removing 2006.1, renaming 2007.1, and pondering 2007.0's fate.
--Kumba
--
Gentoo/MIPS Team Lead
"Such is oft the course of deeds that move the wheels of the world: small hands
do them because they must, while the eyes of the great are elsewhere." --Elrond
--
gentoo-dev@lists.gentoo.org mailing list
